TimeSeries Arrays in Julialang

General Usage
1

I am just starting to develop some time series analysis in Julia, and I was trying to understand which time series array format is the current standard or the most used? I know this is a tough question, but let me try and clarify. I started just checking out TimeSeries.jl, since it is listed on the JuliaStats page. I really wanted a library to handle indexing timestamps and then perform lags or first, second differences on the data.

TimeSeries.jl seems to have these features. But then I was trying some functions from HypothesisTests.jl on the the TimeArray, and there does not seem to be an implementation for TimeArrays in that package. That point alone does not mean anything, but it just made me wonder if people are using TimeArrays versus another array library.

Hence, I was wondering if DataFrames.jl was a more appropriate or common format for Time Series data lately?

2 Likes
2

TimeSeries.jl is a bit cumbersome and lacking on some features. However, seems more lightweight and extendable for time-related functionality than DataFrames… Personally, I prefer Array-like structures to DataFrames…

3

@zgornel thanks for the info. Yeah, I was thinking along the same lines. TimeSeries.jl TimeArrays are nice when your array is already set and you don’t want to do any more manipulations–since the TimeArrays are immutable. So lately my thinking is to convert to TimeArrays as the last step, before estimation. That way, I can use the differencing and lag operators, etc., and don’t need to re-write those things myself. But yes, I would prefer using a regular array structure instead of DataFrames if possible–though there seems to be a nice convertor back and forth from DataFrames and TimeArrays. Thanks again for your input.